International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
Selection of Capacitance for Compensated Self
Excited Induction Generator Using Meta-Heuristic
Approach
Radwan M. AL- Bouthigy
Electrical Engineering Department, Faculty of Engineering, Sana'aUniversity, Yemen
Email address: r.albouthigy@eng-su.academy
Abstract— The performance of self-excited induction generator was
thoroughly analyzed regarding the values of the critical excitation
capacitance and the frequency of the generated voltage. Analytical
expressions were derived for the value of the minimum excitation
capacitor under different load types/operating conditions. The
advised expressions were verified experimentally and by, a robust
meta-heuristic optimization technique, particle Swarm Algorithm.
Furthermore, the feasibility, suitability and effectiveness. of the
proposed formulas were corroborated for generators with different
sizes/ratings through the correlation of the results from the derived
expressions and Particle Swarm Algorithm for such machines.
and rotor currents for given values of capacitance, load and
speed, while the second is to determine the required excitation
capacitance for the desired voltage at given load and speed
levels [7-21].
Extensive research efforts were draft in the past decades
[5-11] for investigating the steady state performance of Self
Excited IG (SEIG) using the steady-state equivalent circuit.
For example, [1] exposes in Section 19.4 a second order slip
equation. It was obtained from the steady-state equivalent
circuit. This equation, as claimed in [1], could predict the
occurrence of self-excitation in SEIG through examining the
roots of the advised equation. However, the numerical values
for all the SEIG parameters including resistance, generated
frequency, excitation capacitive, magnetizing and leakage
reactance’s should be supplied to solve this equation. Thus,
this equation could not identify explicitly, which parameters
affect self-excitation process. Moreover, [1] introduces no
validation for such equation either from simulation or
experimental work.
A mathematical model in [8], [9] was developed for
obtaining the steady state performance of SEIG using the
equivalent circuit. In this approach, the complex impedance is
segregated into real and imaginary parts. The resulted
nonlinear equations are arranged for unknown variables such
as magnetizing reactance (Xm) and frequency (F), while the
remaining machine parameters and operating variables are
assumed constants. Numerical techniques as Newton Raphson
were employed for solving the equations. This approach,
however, requires sophisticated computation capabilities in
terms of speed and storage. In [10] an approach for computing
steady state performance of the self-excited induction
generator is proposed; 4th order polynomial is derived from
the loop equation of the equivalent circuit of the machine. The
roots of this polynomial are determined to check occurrence of
self-excitation and to get the corresponding value of
magnetizing reactance. The approach proposed in [10] has the
advantage of predicting the performance of the machine for
given capacitance/load/speed level. However, the load
considered in this approach is pure resistive, which has less
practical significance.
A mathematical formula was proposed in [11] based on
steady state equivalent circuit of the machine for computing
the minimum value of the capacitance required for selfexcitation and the threshold speed, below which selfexcitation could not be established. Another mathematical
Keywords— Self Excitation, induction generator, particle swarm
algorithm, critical capacitance, magnetizing reactance.
I. INTRODUCTION
Induction Generator (IG) is the preferred option for harnessing
electrical power from non-conventional energy sources,
particularly wind. This is attributed to the salient features of
IG such as: robustness, maintenance free, short- circuit
protection capability and absence of separate DC excitation
system [1-3]. IG could be operated either grid-connected or
off-line; for the case of grid-connected, the reactive power
requirements for maintaining constant voltage at generator
terminals under different load/speed conditions are supplied
by the grid. However, for the case of stand-alone operation,
which is the case for remote and rural locations, the capacitive
excitation is indispensable to regulate the voltage across the
machine terminals [3-7].
The machine terminal voltage decreases/increases with the
load increase/reduction for fixed excitation capacitance and
constant speed operation. To regulate the terminal voltage, the
excitation capacitance has to vary simultaneously with the
load. This is a costly and complicated solution. However, if
the terminal voltage is allowed to vary within a narrow range,
an attractive, in-expensive and simple approach is to use
stepped switched capacitors with the possibility of switching
them on/off with the loads [3-5].
The principle of self-excitation was employed in other
research areas as dynamic braking of three-phase induction
motor; therefore, the techniques for analyzing the behavior of
such machines are of significant practical interest [3], [5]. In
general, there are two scenarios for analyzing the steady-state
performance of self-excited cage induction generator. The first
scenario is to determine terminal voltage, output power, stator
20
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
formula is proposed in [12] for computing the static
performance of the induction generator under wide range of
operating conditions. In [13-15], the performance of separateexcited induction generator is investigated to evaluate the
range of different parameters as voltage, speed and excitation
capacitance, within which self-excitation is possible.
In [16] an analysis for the performance of SEIG based on
eigenvalues is proposed. The lower and upper limits of the
excitation capacitance are identified. However, this approach
is a numerical iterative procedure. The computation time and
speed rely on the initial guess, which is inconvenient.
Moreover, the distinction between the value of minimum and
maximum capacitance is only the value of the initial guess,
either it is small or large. This limits the proposed method
applicability for machines with specific rating [16]. The
critical capacitance under different load conditions was
predicted using simulated annealing optimization technique in
[17]. However, the rotor speed is restricted within a narrow
range. The relation between the excitation capacitance and the
machine parameters and load types could not be visualized.
Moreover, there is no experimental collaboration for the
analysis in [17].
Most of the reported approaches for evaluation of the
steady-state performance of self-excited cage induction
generators require splitting the equivalent impedance into real
and imaginary components. Moreover, the model becomes
rather complicated, if the core losses are included.
Accordingly, several assumptions are taken to simplify the
analysis. Furthermore, different models are used for modeling
the machine with different types of loads/excitation capacitor
arrangements. The coefficients of the mathematical models
vary, which complicate the problem even further. Moreover,
the derived mathematical formulas are only obtained for less
practical loading condition, as resistive loads [8-21].
In this paper, analytical expressions were derived for the
value of the critical capacitance under different load
types/speed levels. Moreover, the value of the frequency of the
generated voltage under different operating conditions is
identified. The factors that affected the frequency of the
generated voltage are addressed. The advantages of the
derived analytical expressions here in this work are:
1. Identifying explicitly the machine parameters that affect
the values of the excitation capacitor and generated
frequency compared with the blind answers of the
numerical methods.
2. Applicability for an arbitrary induction machine
irrespective to rating, while numerical techniques are
usually applied for a specific machine.
3. Addressing different operating/loading conditions as opencircuit, resistive and inductive loads, which as far as we
know were not reported.
are taken as independent variables. The upper/lower limits of
the unknown variables are selected to achieve practically
acceptable values. The results from PSO technique are used
for verifying the derived analytical expression under different
load/speed levels. Furthermore, the derived expressions are
experimentally collaborated.
II. CRITICAL EXCITATION CAPACITANCE
The following analysis is valid for SEIG, whether squirrelcage or wound rotor, provided that the self-excitation arrangements
are allocated in the stator side. The equivalent circuit of SEIG is
shown in figure 1.
Fig. 1. Equivalent circuit of self-excited induction generator
Rs, Rr, Xls, Xlr, Xm, Xc, RL and XL are per phase stator
resistance, rotor resistance, stator leakage reactance, rotor
leakage reactance, magnetizing reactance, excitation
capacitive reactance, load resistance and reactance
respectively. The reactances in the equivalent circuit are
calculated at base frequency. F and υ are p.u. frequency and
speed. Is, Ir, Im, IC and IL are per phase stator, rotor,
magnetizing, capacitor and load currents respectively. The
resistance, reactance, current and voltage of the rotor are
referred to the stator. Applying loop-impedance method [14]
in the equivalent circuit, Figure 1, the following equation
results,
ZTIS=0
(1)
under steady-state self-excitation, Is # 0, thus ZT in (1) is equal
to zero. ZT is given by,
ZT =ZC ZL Z M +ZR+ZS Z M +ZR Z L +ZC+ZMZRZ L +ZC (2)
Where Zs, Zr, Zm and Zc are stator, rotor, magnetizing,
excitation capacitance impedances respectively; they are given
in Appendix. Equation (2) is used for investigating different
load conditions in the following sections.
A. No-Load ZL =
The no-load operation of the SEIG is stimulated by
equating load impedance by infinity ZL =. Substituting the
values of
Zs, Zr, Zm, ZL and Zc into (2) and equating the real and
imaginary parts by zero; the value of the critical capacitive
reactance is obtained for the no-load condition as,
F 2  F  v
R 
R
X c  critical   s  r   X 1  X m 
(3)
Rr
 F F v 
and the frequency of the generated voltage is obtained from,
A meta-heurtics technique, Particle Swarm Algorithm
(PSO) is employed for verifying the derived analytical
expressions. The PSO uses a generic mathematic model for
SEIG; this model is used for any load type/excitation capacitor
arrangement. In PSO, the complex impedance is formulated as
the objective function; and capacitive reactance and frequency
21
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
2X X
1
m
 X 12  F 3  v  2 X 1 X m  X 12  F 2
  X c  X m  X 1   Rs Rr  F  vX c  X m  X 1   0
(4)
X c  critical
In no-load operation of SEIG, the machine slip is almost
zero, and the pu frequency of the generated voltage is nearly
equal to the pu rotor speed. Thus, substituting Fv and setting
F-v 0 in (3), the critical capacitive reactance could be
(5)
X c critical   X1  X m  v 2
(8)
and the frequency of the generated voltage is obtained from,
RL  2 X 1 X m  X 12  X L  2 X 1 X m  X 12  F 3


v RL  2 X 1 X m  X 1   XRs  X m  X 1  F 2
(9)
  X c  X m  X 1  RL  Rs  Rr   Rs Rr RL  X c Rr X L  F
Equation (5) indicates that minimum excitation
capacitance is inversely proportional with square of rotor
speed, leakage and magnetizing reactance’s. Thus, the
excitation capacitance has to fulfill the machine reactive
power requirement for successful operation (5). Equation (5)
coincides with conclusions in Section 19.6 in [1]. Equation (4)
has three roots; two of these roots are discarded due to their
extraordinary values. Thus, third root depicts the frequency.
The frequency of the generated voltage was found to be
independent of the leakage reactance, and it is speed
dependent.
R
B. Resistive load Z L  L
F
The critical capacitance for the resistive load case could be
R
obtained by equating load impedance by Z L  L ; then
F
Substituting the values of Zs, Zr, Zm , ZLand Zc into (2) and
equating the real and imaginary parts by zero. The value of the
critical capacitive reactance for resistive load case is given by,
R

Rr
RL  s2 
  X1  X m 
F  F  v  
F

X c  critical 
(6)
2 X 1 X m  X 12
Rr R
Rr R

F 3  F  v F 3  F  v
F2
vX c  X m  X 1  RL  Rs   0
Equation (8) depicts clearly the relation between the
excitation capacitance and load reactive power requirements,
which are represented in load inductive reactance XL. Again,
the critical capacitive reactance varies with the prime mover
speed and leakage and magnetizing reactance. Equation (9)
has three roots similarly to (4) and (7), only one of these roots
has acceptable value. Accordingly, the frequency of the
generated voltage is found to be independent on leakage
reactance; however it varies with the load, speed and motor
copper losses.
III. PERFORMANCE OPTIMIZATION
For optimizing the performance of SEIG regarding the
minimum capacitance and the corresponding generated
frequency for given speed and load levels, the problem is
mathematically formulated as,
Minmize Z  F , X c 
(10)
Usually the frequency and capacitive reactance are
bounded to reduce the computation time. This approach is
used to verify the analytical results in (3)-(9). To fulfill (10),
ZT has to be segregated into real and imaginary parts;
accordingly two nonlinear equations are obtained with Xc and
F as unknown variables,
f  F , X c   a1 F 3  a2 F   a3 X c  a4  F   a5 X c 
(11)
and the frequency of the generated voltage is obtained from,
RL  2 X 1 X m  X 12  F 3  vRL  2 X 1 X m  X 12  F 2
  X c  X m  X 1  RL  Rs  Rr   Rs Rr RL  F
R

Rs Rr X L
Rr
2
RL  s2 
  X1  X m    2 X1 X m  X1  X L 
F  F  v  
F  F 
F


2 X 1 X m  X 12 X L  X 1  X m 
Rr R
Rr R



F 3  F  v F 3  F 
F2
F2
g  F , X c   b1 F 4  b2 F 3   b3 X c  b4  F 2
(7)
vX c  X m  X 1  RL  Rs   0
Equation (6) indicates that the critical capacitance is load
dependent. Again, it varies with rotor speed and leakage and
magnetizing reactance’s. Similarly to (4), equation (7) has
three roots; two of them are ignored, while the third provides
the frequency of the generated voltage. The frequency is found
to be a function in load resistance and machine speed, which
concurs with the slip equation advised in [1].
R
C. Inductive Load Z L  L  jX
F
Inductive load represents the generic case for the SEIG.
R
The load impedance is given by, Z L  L  jX Substituting
F
the values of Zs, Zr, Zm, ZL and Zc into (2) and equating the real
and imaginary parts by zero; the value of the critical
capacitive reactance for inductive load,
  X c b5  b6  F   b7 X c 
(12)
where the coefficients a1-a5 and b1-b7 are given in the Appendix .
A. Particle Swarm Algorithms
The PSO method is a member of wide category of Swarm
Intelligence methods for solving the optimization problems. It
is a population based search algorithm where each individual
is referred to as particle and represents a candidate solution.
Each particle in PSO flies through the search space with an
adaptable velocity that is dynamically modified according to
its own flying experience and also the flying experience of the
other particles [22]. The principle and mechanism of PSO are
adequately covered in optimization references as [22]. The
PSO was selected here in this work due to its relevant
features, which are:
1. The capability of searching many points in the search
space simultaneously,
2. The independency on the starting point(the initial solution
is generated randomly),
22
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
3. Minimum requirements, as it needs only objective function
and fitness function to modify and direct its performance,
4. Stochastic search method, which implies that the
probability contributes to the most of the decisions.
IV.
implemented as a Matlab function that accepts the machine
and load parameters, and it returns a minimum value of the
amplitude of ZT at a given frequency and capacitive reactance.
Moreover, the analytical formulas were corroborated
experimentally.
The steady-state performance of machine under concern was
evaluated for different levels of excitation capacitance in the
following sections.
STUDY CASE
A delta connected 415V, 3.7kW SEIG is used for
validating the derived expressions. The machine parameters
are shown in Table I. Xm, the magnetizing reactance is
determined experimentally through driving the machine at
synchronous speed by a DC motor. The input frequency is
kept at base value; while the input voltage is allowed to vary.
The input impedance per-phase is calculated for different
input voltage, to estimated the value for Xm. The drop in the
stator impedance is taken into consideration in calculating airgap voltage Vg. The magnetizing reactance is shown in the
figure 2.
TABLE I. Parameters of 3.7KW IG
Number of poles
Frequency
Stator resistance Rs (p.u.)
Rotor resistance referred to the stator Rr (p.u)
Stator leakage reactance Xls (p.u)
Rotor resistance referred to the stator Xlr (p.u)
V.
STUDY-STATE PERFORMANCE
To evaluate the steady-state performance of the IG, the
value of magnetizing reactance Xm and the generated voltage
has to be determined for given speed, load and excitation
capacitance. The magnetizing reactance Xm could possibly be
computed under these conditions using PSO, and hence the
mathematical formulation of the problem is given by,
(14)
Minmize Z  F , X m 
The frequency and magnetizing reactance are bounded. To
fulfill (14), ZT again has to be segregated into real and
imaginary parts; accordingly two nonlinear equations are
obtained with Xm and F as unknown variables,
U  X m , F   F  c1 X m  c2   F 2  c3 X m  c4 
(15)
 F  c5 X m  c6    c7 X m  c8 
4
50Hz
0.053 p.u
0.061 p.u
0.087 p.u
0.087 p.u
W  X m , F   F  d1 X m  d 2   F 3  d3 X m  d 4 
 F 2  d5 X m  d 6   F  d 7 X m  d8   d 9
where the coefficients c1-c8 and d1-d9 are given in the
Appendix .
To determine the minimum Xm for the given condition, a
program coding PSO is written in Matlab environment.
After determining the value of magnetizing reactance Xm and
frequency F for given capacitance, speed, and load, the
generated voltage Vg could be obtained from figure 2. Then,
the load voltage, current and power could be obtained
using the equivalent circuit by,

V / F  * Z c
IL 
 Z s  Z L / / Z c  Z c  Z L  

(17)
VL  Z L * I L


2
PL  m I L RL


where the number of phases m for the machine under concern
is 3.
Fig. 2. Air gap voltage-frequency ratio (Vg/F) versus the magnetizing
reactance Xm, experimental data (asterisk), fitted curve (solid), linear
approximation (dashed)
VI. RESULTS AND DISCUSSIONS
The saturated magnetizing reactance is approximate by,
Vg
F
 1.62  0.32 X m
(16)
The variation of the minimum capacitance and generated
frequency with rotor speed for different operating scenarios
are shown in figure 3 and figure 4 respectively. A good
correlation is shown in figure 3 between the results from testring, optimization technique and equations (3)-(8). The values
of threshold excitation capacitance from test-ring are slightly
higher than those from PSO and analytical expressions. This
attributed to the leakage, harmonics and parasitic inductances.
Figure 3 shows that the critical capacitance for inductive load
is significantly higher than that of resistive load and no-load.
(13)
This approximation as shown in figure 2 correlates with
the experimentally and the fitted data. To corroborate the
derived expressions (3)-(8), a program is written in Matlab Mcode for implementing PSO using PSO toolbox fitted in
Matlab. The maximum number of generation is limited to 20,
as compromise between accuracy and computation time. The
objective function for PSO is obtained from (11) and (12). It is
23
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
This is attributed to the fact the excitation capacitor has to
satisfy the reactive power requirements for the load and the
generator simultaneously. Accordingly, capacitive load may
require less capacitance than no-load case. The results in
figure 3 coincide with conclusion in Section 19.6 in [1],
particularly for no-load condition.
slip at maximum developed power is approximately 0.35;
which could explain the raised values of slip at other operating
points. In case of no-load, the generated frequency almost is
rotor velocity dependent.
The dependency of the generated voltage on excitation
capacitor is exploited in figure 5 for three levels of rotor
speed. The generated voltage shown in figure 5 is computed
using steady-state equations and measured from the test-rig. In
the test-rig, the terminal voltage and stator current are
measured; then the generated voltage is obtained by
subtracting the drop in stator impedance from the measured
terminal voltage.
Fig. 3. Critical capacitance versus rotor speed for No-load (dots), resistive
load at rated power (stars), inductive load at rated power and 0.8pf lag
(squares) from analytical expressions(dashed line),PSO(solid line) and
experimental (no line) for 3.7kW machine
The critical capacitance is a speed dependent, figure 3; the
capacitance drops nearly by 40% for 25% increase in the
speed. In figure 3, the deviation between the results from
optimization and from the derived equations is attributed to
the limits imposed on independent variables.
Fig. 5. Generated voltage versus capacitance at no load at 0.8pu rotor velocity
(dots), 1pu rotor velocity (stars), 1.2pu rotor velocity (squares), calculation
(solid line) and experimental (no line)
Again good corroboration between the computed and
experimental results is shown in figure 5. The minimum
capacitance required to ensure successful build up increases as
the speed decreases. The generated voltage from the test-rig is
slightly lower than the computed. This is attributed to
saturation, core losses and drop in stator circuit. Figure 3
shows that for 3.7kW, 15µf is the minimum value of the
excitation capacitance at rated speed, below which the selfexcitation is not possible. Moreover, the Figure shows that
there is upper limit for the excitation capacitor above which
the machine reverts into saturation. In the saturation the
increase in the excitation capacitance will not produce
significant increase in output power/generated voltage, which
could not overwhelm the increase in capacitor size, cost and
losses.
The variations of the terminal voltage with output power
for different levels of excitation capacitance are shown in
figure 6 for resistive and inductive load at 0.8 pf lag types.
The terminal voltage/output power of IG increases/decreases
with increase/decrease in the excitation capacitance, figure 6,
provided that saturation is not reached. Figure 6 shows that the
voltage regulation for the inductive load is inferior compared
to that of the resistive load; this may be attributed to the
function of the excitation capacitor in case of inductive load in
fulfilling the reactive power requirements of the load and the
generator. It is observed from figure 6, that the characteristic
Fig. 4. Generated frequency versus rotor speed for No-load (dots), resistive
load at rated power (stars), inductive load at rated power and 0.8pf lag
(squares) from analytical expressions (dashed line), PSO(solid line) and
experimental (no line)
The generated frequency from analytical expressions
generally correlates with those from PSO and test-rig, figure
4; however, the results from optimizations for inductive load
deviate from experimental and derived equations. This is
attributed to the limits imposed on the PSO. The machine
under consideration has large rotor resistance, Table I; thus the
24
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
of self-excited induction generator is nearly similar to that of
separately-excited DC generator.
shown in figure 8 and figure 9. The parameters of these
machines are given in the Appendix in Tables III and IV for
1.1kW and 373kW respectively. The magnetization curves (BH) for these machines are given in references [2] for 373kW and
[16] for 1.1kW. It is worth to mention that magnetizing
reactance’s used for developing figure 3, figure 8 and figure 9
are corresponding to edge between saturation and un-saturation
zones in (B-H) curves. This is to simplify the analysis.
However, for investigating the steady-state performance of
3.7kW, values of magnetizing reactance are obtained from Fig.
2 and from optimization. The critical excitation capacitance for
1.1kW and 373kW are computed for wide range of speed from
0.7 to 1.4pu, as in figure 3.
Fig. 6. Terminal voltage versus output power at rated speed and 29.6µf
capacitor for pure resistive load: calculated (solid), experimental (dots);
inductive load, 0.8pf lag calculated (dashed), experimental (stars)
The dependency of the output power on the excitation
capacitance is exploited in figure 7, where the output powers
are plotted versus the capacitance for constant terminal
voltage/speed.
Fig. 8. Critical capacitance with rotor speed for No-load (dots), resistive load
at rated power (stars), inductive load at rated power and 0.8pf lag (squares)
from analytical expressions (dashed line), PSO solid line) for 1.1kW machine
Again good correlation between results from the advised
formulas and PSO is illustrated in Fig. 8, particularly for the
no- load and resistive load scenarios. Comparing Figs. 3 and 8
reveals that excitation capacitance varies noticeably with
magnetizing reactance. Moreover, as mentioned before, the
minimum capacitance could be considered as speed
dependent, figure 3 and 8. It is worth to mention that the
deviation between the results from PSO and the derived
equations is attributed to the constraints on independent
variables.
Comparing figure 8 and figure 4 in ref. [16] validates the
derived expressions here; as high degree of resemblance is
obvious between these two figures. To expose the similarity of
the results from the proposed formulas here and those from
numerical techniques in [16], a number of minimum
capacitance values are given in table II.
Table II shows that the results have significant correlation
for no-load case. For resistive and inductive loads, the derived
expressions here give slightly small regarding to [16]. This
may be attributed to the value of the magnetizing reactance.
As mentioned before that ref. [16] utilize a numerically
interactive method based on eigenvalues, which could not
identified explicitly the influence of different machine and
load parameters on the value of the critical capacitance. In
the contrary the expressions proposed here, equations (3),
(6), and (8) ,show explicitly the role of different parameters
Fig. 7. Excitation capacitance versus output power at rated terminal voltage
and rated speed
For constant terminal voltage/speed, the capacitance has to
increase for an increase in the output power. The excitation
capacitance in figure 4 and figure 5 is limited to 28µf. This is
to avoid the operation in saturation. Figure 7 illustrates that the
excitation capacitance has to increase by around 60% for 50%
increase in the load power. Again, the results in figure 7
concur with the conclusions in Section 19.6 in [1].
VII. VERSATILITY OF THE DERIVED FORMULAS
The ability of proposed expressions to determine the critical
excitation capacitance for other machines was investigated. The
critical excitation capacitance for a number of induction
generators with different power levels are estimated by the
advised formulas and PSO Two examples are given below; the
minimum capacitance for 1.1kW, [16] and 373kW, [2], are
25
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
in the capacitor value. This is useful tool for the design and
the operation of the SEIG.
Speed
Load
0.8pu
1pu
1pu
1.1pu
1pu
No-load
No-load
1pu
resistive
1pu
resistive
1pu at
0.8pf
2. Capacitor banks are mandatory for stand-alone operation
TABLE II. Critical excitation capaitance
Minimum capacitance from derived Minimum capacitance
expressions here
from [16]
78μF
79μF
48.3μF
49.5 μF
94μF
112μF
79μF
98μF
130μF
136μF
3.
TABLE III. Parameters of 1.1KW IG [16]
Line voltage
220V
Number of poles
2
4.
5.
Frequency
60Hz
Stator resistance Rs (p.u.)
0.0946 p.u
Rotor resistance referred to the stator Rr (p.u) 0.0439 p.u
Stator leakage reactance Xls (p.u)
0.0865 p.u
Rotor resistance referred to the stator Xlr (p.u) 0.0865 p.u
6.
7.
8.
TABLE IV. Parameters of 373KW machine [2]
Line voltage
2300
Number of poles
2
Frequency
50Hz
Stator resistance Rs (p.u.)
0.0185p.u
Rotor resistance referred to the stator Rr (p.u) 0.0132 p.u
Stator leakage reactance Xls (p.u)
0.0850 p.u
Rotor resistance referred to the stator Xlr (p.u) 0.0850 p.u
9.
of SEIG for supplying the machine with reactive power
requirements, (5).
Analytical expressions were derived for minimum
capacitance and generated frequency for different load
types. These formulas show explicitly the parameters that
affected the critical capacitance and the generated
frequency. The minimum capacitor varies nearly inversely
with the square of the rotor speed. The generated
frequency drops with the load increase, while excitation
capacitance increases.
A good correlation between the results from the derived
expressions, PSO and the test-rig.
The terminal voltage of IG increases/decreases with
increase/reduction in the output capacitance
For constant terminal voltage and speed, the excitation
capacitance has to vary with the load.
The performance of self-excited induction generator is
investigated here for different loads/operating conditions,
which is likely novel.
The advised expressions are validated for generators with
different sizes/ratings by comparing the results with those
obtained from PSO for such machines. This confirms the
effectiveness, robustness and feasibility of these formulas.
The proposed analytical expressions here and the presented
results concur with the reported in literature, as in [1] and
[16].
APPENDIX
Rs
X
R
 jX L , Z r  r  jX lr , Z m  jX m , Z c   j 2c
F
F
F v
RL
 jX L
and Z L 
F
The coefficients of equation (11)
a1 = - ((2Xm+X1) X1 RL+XL (X1+Xm) (Rr+Rs))
a2= (2Xm+X1) X1 RL υ+Rs XL (X1+Xm) υ
a3= (RL + Rs +Rr) (Xm+X1) + (XL RL), a4= Rs RL Rr
a5= - (Rs+RL) (Xm+X1) υ
The coefficients of equation (12)
b1= -X1 XL (X1+2Xm), b2= - X1 XL (X1+2Xm) υ
b3= (Xm+X1) (XL+X1)+(X1Xm)
b4= Rs XL Rr+RL (Xm+X1) (Rs+Rr)
b5= - ((Xm+X1) (XL+X1) +X1 Xm) υ
b6= -Rs RL (Xm+X1) υ, b7= - Rr (Rs+RL)
The coefficients of equation (15)
c1=-XL (Rs+Rr) - (2X1) RL, c2=-XLX1 (Rs+Rr) - (X2) R1 L
c3= (2X1RL υ + RsXL υ), c4=X1(RsXL υ +X1RL υ)
c5=Xc(RL+Rs+Rr), c6=XLXcRr+RsRLRr+X1Xc (RL+Rs+Rr)
c7=-Xc (RL+Rs) υ, c8=-X1Xc (RL+Rs) υ;
The coefficients of equation (16)
d1=-2X1XL, d2 =-XL (X1 2), d3 =2X1XLv
d4 =(X1 2)XLv, d5 =(Rs RL + 2X1Xc + XL Xc + Rr RL)
d6=XL (RsRr+XcX1) + (X1RL) (Rr+Rs) +(X2) Xc
d7= υ ((-2X1Xc) - (RsRL) - (XcXL))
d8=X1 υ (-RsRL-X1Xc-XcXL), d9=-XcRr (Rs+RL)
Zs 
Fig. 10. Capacitive reactance (pu) for 373kW (dotted), 3.7kW (asterisks) and 1.1kW
(squares) for full load at 0.707pf lag versus rotor velocity (pu)
Figure 10 shows that for full load at 0.707 pf lag, the
373kW machine has highest pu capacitive reactance over the
entire speed range; while 1.1kW has the lowest pu captive
reactance. This is possibly attributed to the limits imposed
in generating frequency.
VIII. CONCLUSION
The following conclusions can be drawn:
1. The IG seems to be the preferred option for harvesting
electrical power from renewable energy sources, due to its
salient advantages: robustness, maintenance free and
reduced volumetric dimension/cost.
26
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.
International Research Journal of Advanced Engineering and Science
ISSN: 2455-9024
[13] N. H. Malik, A. H. AL- Bahrani, “Influence of terminal capacitor on the
performance characteristics of a self – excited induction generator,”
IEE, vol. 137, pp.168 -173, 1990.
[14] S. P. Singh, B. Singh, and M. P. Jain, “Performance characteristics and
optimum utilization of a cage machine as capacitor excited induction
generator,” IEEE Transactions on Energy Power Conversion, vol. 5, pp.
679–685, 1990.
[15] M. H. Haque, “A novel method of evaluating performance
characteristics of a self-excited induction generator,” IEEE Transactions
on Energy Power Conversion, vol. 24, pp. 358–365, 2009.
[16] L. Wang and C. Huei, “A novel analysis on the performance of an
isolated self-excited induction generator,” IEEE Transactions on Energy
Power Conversion, vol. 12, no. 2, pp. 109-117, 1997.
[17] R. Ali, A. A. AbdElhafez, and G. El Sady, “Performance optimization of
self-excited induction generator using simulated annealing technique,”
Journal of Electrical Engineering, JEE, vol. 1, 2012
[18] S. N., Mahato, P. S. Singh, and M. P. Sharma, “Capacitors required for
maximum power of a self-excited single-phase induction generator
using a three-phase machine,” IEEE Transactions on Energy Conversion
, vol. 23, no. 2, pp. 372-381, 2008.
[19] M. H. Haque and A. I. Maswood, “ Determination of excitation
capacitance of a three-phase self-excited induction generator,” IEEE
Power and Energy Society General Meeting, pp. 1-6, 2012.
[20] S. P. Singh, B. Singh, and M. P. Jain, “A new technique for the analysis
of excited induction generator,” IEEE Transactions on Energy Power
Conversion, vol. 23, pp. 647–656, 1993.
[21] N. H. Malik and S. E. Haque, “Steady state analysis and performance of
an isolated self-excited induction generator, “IEEE Transactions on
Energy Conversion, vol. EC-1, no. 3, pp. 134-140, 1986.
[22] S. R. Singiresu, Engineering Optimization Theory and Practice, New
Jersey: John Wiley & Sons, pp. 708, 2009.
REFERENCES
[1]
Bodlea and S. A. Nassar, Induction machine handbook, Second Edition,
CRC press, 2010.
[2] P. C. Krause, O. Wasynczuk, and S. D. Sudhoff, Analysis of Electric
Machinery, IEEE Press, 1994.
[3] R. C. Bansal, “Three-phase self-excited induction generators: an
overview,” IEEE Transaction on Energy Conversion, vol. 20, no. 2, pp.
292-299, 2005.
[4] S. Boora, “On-set theory of self-excitation in induction generator,”
International Journal of Recent Trends in Engineering, vol. 2, no. 5, pp.
213-219, 2009.
[5] S. Mahley, “Steady state analysis of three-phase self-excited induction
generator,” M. Sc. Thesis, Thapar University, Patiala, 2008.
[6] H. Amimeur, R. Abdessemed, and E. Merabet, “Modeling and analysis
of dual-stator windings self-excited induction generator,” Journal of
Electrical Engineering (JEE), vol. 8, no. 3, pp. 18-24, 2008.
[7] Y. J. Wang and S. Y. Huang, “Analysis of a self-excited induction
generator supplying unbalanced load,” International Conference on
Power System Technology, vol. 2, pp. 1457-1462, 2004.
[8] S. S. Murthy, O. P. Malik, and A. K. Tandon, “Analysis of self excited
induction generators,” Proceeding of IEE, vol. 129, pp. 260-265, 1982.
[9] N. H. Malik and A. A. Mazi, “Capacitance requirements for isolated
self excited induction generators,” IEEE Transactions on Energy
Conversion, vol. EC-2, pp. 62-69, 1987.
[10] A. K. Tandon, S. S. Murthy, and G. J. Berg, “Steady state analysis of
capacitor self excited induction generators,” IEEE Transactions on
Power Apparatus and Systems, vol. PAS-103, no. 3, pp. 12-618, 1984.
[11] A. K. AL- Jabri and A. I. Alolah, “Capacitance requirement for isolated
self – excited induction generator,” Proceeding IEE, vol. 137, pp. 154-159,
1990.
[12] A. K. AL- Jabri and A. I. Alolah, “Limits on the performance of three
self excited induction generators,” IEEE Transactions on Energy Power
Conversion, vol. 5, pp. 350–356, 1990.
27
Radwan M. AL- Bouthigy, “Selection of capacitance for compensated self excited induction generator using meta-heuristic approach,”
International Research Journal of Advanced Engineering and Science, Volume 1, Issue 2, pp. 20-27, 2016.